Constant 95% to100% CPU Usage

[MBBNET]

We have over 60 Netonix switches in our network with several of them showing between 90% and 100% CPU usage consistently. What would cause this and how would one fix this issue. We are running 1.4.3rc7 on some and 1.4.4 on the others. Each showing the same result.

[sirhc]

First please post up a screen grab of the Device/Status Tab. You upload pictures to your post using the Upload attachment link under the BLUE SUBMIT button and then click place in line where you want it to be and it inserts the picture where your cursor currently is in the post your creating

Next go to the Device/Console Tab

Enter/type the command "cmd" and press Enter to drop to the Linux Shell level

Now run the command "top" and press enter and post a screen capture so we can see what services are running.

Hit Print Screen to capture your screen

CTRL+Z will stop the top command

Type "exit" and hit Enter to return to normal CLI

Next past the screen grab into MS Paint then crop it and save as a JPEG so you can upload the picture to your post.

Services that could cause this are mis-configured SMTP service, Discovery Protocols, mis-configured TFTP backup, or Discovery Tab. But I thought these were all resolved?

You can try disabling these services then re-enable them one at a time to determine which one might be your issue.

How large of a flat network segment are you running?

[MBBNET]

Hear is a screen capture of the unit in question. As far as that segment network, it connects to our core router and handles the distribution there. We replaced the WS-12-250A with the WS-12-400 last week and still the problem is there. We are replacing it with the WS-24-400 today or tomorrow as we need more ports anyway. Thanks so much for looking at this for us...

[sirhc]

I am confused, the CPU is showing 47% in your picture not 90%+ which is normal?????

Also please post up the Device/Status Tab

Also when you first log into a switch a CPU will Spike for awhile as it is busy sending data to your browser but should settle down to 50% or less as a few seconds.

The CPU has nothing to do with capacity of the switch as the CPU your looking at only drives the UI/CLI and runs control Daemons like RSTP, SMTP, LAG, STATS, SNMP, and such which is why you do not want to see it at 90%-100% as it could affect those daemons and cause issues.

Also sometimes after after upgrading to v1.4.4 if have high CPU or something acts weird you should reboot it once more as the previous reboot is handled by the previous firmware version.

Also as I said if after rebooting your CPU does not settle down try disabling the features I suggested in my last post and reboot and let me know.

[MBBNET]

I will do what you suggest and see how things settle down. Here is the screen grab from the "Status" page I was referring to.

[sirhc]

Was the Top screen grab and the Device/Console/Top screen grab from the same switch?

Also the Device/Status grab was from v1.4.4, please upgrade to v1.4.4 then reboot after upgrade then try disabling the services I suggested and possibly reboot after they are disabled.

[MBBNET]

Yes they are the same switch. I went ahead and upgraded to the 1.4.4 firmware on it and there were no changes. I am not sure how to disable the services you spoke of but will look into this as well.

[sirhc]

Yes they are the same switch. I went ahead and upgraded to the 1.4.4 firmware on it and there were no changes. I am not sure how to disable the services you spoke of but will look into this as well.

To disable those services you got to the Device/Configuration Tab

Disable SMTP Alerts
Disable Discovery Protocols such as UBNT, LLDP, CDP, disable TFTP Backup, and the Discovery Tab

AFter you disable all of them and save and apply and if the CPU does not calm down then tell it to reboot.

If that does nothing then Load up Teamviewer on your computer and send me a PM with your cell #

[roman.truog@ict-plus.ch]

Good day, we have the same problem in one of our networks. The utilization of the responsible process can not be reconstructed via the linuz shell.

[sirhc]

Well obviously this is not normally experienced by most users so we need to find out what service it is.

As described above in this this thread:

Have you tried disabling the following services and then rebooting:
Discovery Protocols - UBNT, CDP, LLDP
Discovery TAB
SMTP email notification
SNMP

Major CPU can also be caused by:
Having 2 Netonix Manager Installs monitoring the same switch.
Using SNMP and polling the switch too often. I have see some people try to poll the switch every second which is INSANE.
Having multiple logins to the UI/CLI as ever connection uses CPU and memory


You need to disable all the things you can until CPU utilization normalizes then enable 1 item at a time and observ to narrow it down.

Since we know that this is not normal behavior it has to be something with your installation. Could you installation be exposing a bug....maybe, or it could be something you need to fix.

But step one is to try and figure out WHAT is causing this then go from there. Be a little bit if a detective and investigate your issue then come to us with what you found.